1238 
(Se considered as bivalent, S as tetravalent) concord well. The other 
substances are “hopelessly aberrant’, as M himself remarks with 
respect to CO. (loc. cit, p. 195). *) 
To redress this he assumes C and O to be monovalent in CO; 
~C and O divalent (mark C divalent) in CO,; S hevavalent in CS, and H,S; 
N monovalent and O bivalent in N,O; N and O both monovalent in 
NO; N and P pentavalent in NH, and PH,; Cl trivalent in SnCl, (!), 
but again monovalent in HCL In GeCl, three atoms Cl would be 
trivalent, but the fourth monovalent!! (loe. eit. p. 259), though M 
also pleads for the inaccuracy of the critical data for GeCl,. 
And to further the good cause CS, NH,, PH,, H,S, HCl are declared 
to be associated at 7, (loc.cit. p. 190 et seq.).*) Then Cl can also 
be taken frvalent in HCl. 
I should not have diseussed all these things at such length, if 
vaN DER Waars had not stated in his paper on the additivity of 
the h-values (cited by us in § 1), that M had found “very remark- 
able relations” with regard to a. I thought it therefore necessary to 
criticize these papers of M. 
d. The carbon hydrogen. Here we find 13 of the 17 substances 
of our table in accordance with M’s rule. But CH,,C,H,,C,H,, C,H, 
(important members, indeed, of the series) deviate greatly. For CH, 
MATHEWS again tries to save his cause by declaring the factor *7/,, 
to be faulty (comp. Footnote 3 on p. 1236)*), and as for C,H,: one 
C-atom would be divalent, the other tetravalent | 
The cause why M’s rule happens to hold good for the higher 
members of the series, is this that the values of Va are simply 
proportional to the number of H-atoms, and that for compounds, in 
which also C,N, and O occur by the side of H, the fundamental 
values of these elements differ little (resp. 3,1, 2,9, 2,7). But when- 
ever Cl occurs, the fundamental value of which is about double the 
value, or S etc, the rule does not hold good at all. 
e. Other organic compounds. As may therefore be expected 
CH,Ci, CHCl,, CCl,, C,H,Cl,, C,H,S deviate more or less considerably. 
1) Instead of VC=1,5>X 10-2 we find for V CX10° resp, 1,4, 0,97, 1,2, 
1,3, 1,8, 5,8, 2, 1,24, 0,94, 1,7, 1,85, and 2. Deviations therefore of on an 
average 459/, (for SnCl, even 2909/9). 
*) M. also asserts that H,Q0 — because it departs more than 700/, from his 
rule — is the most associated (at 7% ) of all substances known. We found H,0 
at 7 only slightly associated, just as C,H;OH ete. Much less than CH30H and 
acetic acid e.g. 
5) M even pronounces the conviction that if only the factor °7/,, were duly cor- 
rected for every substance, all the discrepancies would vanish. Indeed. 
